Robotics Redefine Risk and Liability in Insurance
Physical AI is reshaping risk and liability as intelligent systems move from code to concrete action.
Market growth and early incidents
Morgan Stanley projects the humanoid robotics market could reach $5 trillion by 2050, with close to 1 billion units in operation worldwide. The forecast reflects expanding use in logistics, health care and public spaces, where machines are expected to share tasks with people.
In June 2026, Waymo recalled 3,871 autonomous‑vehicle robotaxis after several reports that the units entered closed freeway construction zones and kept traveling at speed. No injuries were recorded, but the episode showed a growing challenge: software must interpret and react to rapidly changing physical environments.
Related: Two-Thirds of Pension Pots From Growth
The shift from pure software glitches to tangible harm is evident. A coding error that once produced a biased recommendation can now cause a collision, a fire or a production line shutdown. Examples include warehouse automata bumping into staff, delivery drones dropping parcels on pedestrians, and manufacturing devices sparking flames.
U.S. occupational data show 77 incidents involving automated units from 2015‑2022, resulting in 93 documented injuries such than finger amputations and fractures. Those figures are likely to rise as deployment expands.
Compared with the first wave of industrial automation, today’s collaborative systems operate in less controlled settings, which makes risk assessment more complex. Earlier factories could isolate machines behind cages; now bipedal assistants handle open floors alongside workers, increasing the chance of unexpected contact.
Related: Core Data Record 3.3 Boosts Treaty Reinsurance
Liability, regulation and insurance response
When a heavy load falls from a humanoid assistant onto a contractor, responsibility may rest on several parties: the maker if a design flaw exists, the software creator if an algorithm miscalculates balance, the operator if safety protocols are lax, or the service technician if a sensor fails. This overlapping exposure blurs the lines between product, general, professional and cyber coverage.
Regulators are adapting. In Europe, the Product Liability Directive and the AI Act strengthen accountability for AI‑enabled products and other high‑risk systems, including many robotics applications.
Beyond direct damage, the supply chain introduces additional exposures. Specialized inputs such as semiconductors, sensors and high‑energy batteries create geopolitical and logistics risks, prompting demand for cargo and marine cover. Battery handling also raises environmental liability concerns that traditional property policies may not fully address.
